David Wilson

Once #39 is complete, it would become possible to have an index covering multiple collections. This has interesting interactions with implementing joins. It's also another use case where explicitly including...

Presently the unpacked key is incrementally encoded into a brand new PyString, which is then discarded. This is fairly pointless, it should be possible to skip this step entirely

This has two parts: It should be possible to do a single-pass transformation from ticket #41 structs to a JSON representation, eliminating an intermediary representation. This may be especially appealing...

This ticket is to implement a new Struct class that avoids the immediate decoding presently involved in doing reads, to allow dynamic generation of compact encodings in order to support...

Index and collection type signatures (prevent writes using broken configuration)

1

This is mostly about doing everything possible to make it safe to upgrade an existing application. Currently few checks are done on whether runtime state matches the runtime that created...

Implement Collection.putbatch()

1

Some method that allows bulk loading of batch records, without first saving individual records to the store, just to later delete them during `batch()`. Using existing API pattern, `putbatch()` call...

keylib presently capture's a TZ-aware datetime's UTC offset in 15 minute increments, and includes that in the key encoding. The upside of this that for the cost of 4 bits,...

Write high level use case docs

1

Make Dariusz happy

join()/merge() helper functions/classes

1

Once Key creation (issue #23) and iteration (#37) are made cheap, implementing some less trivial query styles becomes interesting. Not decided on what this should look like. Maybe: ``` python...

Fix query styles / string suffix problem

1

Related to ticket #1, need to add at least one more query style, and give much better documentation examples. There is at least the open/closed string suffix issue, which might...